In [1]:
# imports
from src.config import *
from src.mlflow_logging import train_and_log_model
In [2]:
import mlflow
experiments = mlflow.tracking.MlflowClient().list_experiments()
for experiment in experiments:
print(f"Experiment ID: {experiment.experiment_id}, Name: {experiment.name}")
Experiment ID: 0, Name: Default Experiment ID: 1, Name: DecisionTreeClassifier_1 Experiment ID: 10, Name: ExtraTreeClassifier_4 Experiment ID: 11, Name: ExtraTreeClassifier_5 Experiment ID: 12, Name: ExtraTreeClassifier_6 Experiment ID: 13, Name: RandomForestClassifier_1 Experiment ID: 14, Name: RandomForestClassifier_2 Experiment ID: 15, Name: RandomForestClassifier_3 Experiment ID: 16, Name: RandomForestClassifier_4 Experiment ID: 17, Name: RandomForestClassifier_5 Experiment ID: 18, Name: RandomForestClassifier_6 Experiment ID: 19, Name: ExtraTreesClassifier_1 Experiment ID: 2, Name: DecisionTreeClassifier_2 Experiment ID: 20, Name: ExtraTreesClassifier_2 Experiment ID: 21, Name: ExtraTreesClassifier_3 Experiment ID: 22, Name: ExtraTreesClassifier_4 Experiment ID: 23, Name: ExtraTreesClassifier_5 Experiment ID: 24, Name: ExtraTreesClassifier_6 Experiment ID: 25, Name: AdaBoostClassifier_1 Experiment ID: 26, Name: AdaBoostClassifier_2 Experiment ID: 27, Name: AdaBoostClassifier_3 Experiment ID: 28, Name: AdaBoostClassifier_4 Experiment ID: 29, Name: AdaBoostClassifier_5 Experiment ID: 3, Name: DecisionTreeClassifier_3 Experiment ID: 30, Name: AdaBoostClassifier_6 Experiment ID: 31, Name: GradientBoostingClassifier_1 Experiment ID: 32, Name: GradientBoostingClassifier_2 Experiment ID: 33, Name: GradientBoostingClassifier_3 Experiment ID: 34, Name: GradientBoostingClassifier_4 Experiment ID: 35, Name: GradientBoostingClassifier_5 Experiment ID: 36, Name: GradientBoostingClassifier_6 Experiment ID: 37, Name: XGBClassifier_1 Experiment ID: 38, Name: XGBClassifier_2 Experiment ID: 39, Name: XGBClassifier_3 Experiment ID: 4, Name: DecisionTreeClassifier_4 Experiment ID: 40, Name: XGBClassifier_4 Experiment ID: 41, Name: XGBClassifier_5 Experiment ID: 42, Name: XGBClassifier_6 Experiment ID: 43, Name: CatBoostClassifier_1 Experiment ID: 44, Name: CatBoostClassifier_2 Experiment ID: 45, Name: CatBoostClassifier_3 Experiment ID: 46, Name: CatBoostClassifier_4 Experiment ID: 47, Name: CatBoostClassifier_5 Experiment ID: 48, Name: CatBoostClassifier_6 Experiment ID: 5, Name: DecisionTreeClassifier_5 Experiment ID: 6, Name: DecisionTreeClassifier_6 Experiment ID: 7, Name: ExtraTreeClassifier_1 Experiment ID: 8, Name: ExtraTreeClassifier_2 Experiment ID: 9, Name: ExtraTreeClassifier_3
Other algorithms¶
- Linear_model: LogisticRegression
- Naive_bayes: GaussianNB, BernoulliNB
- Neighbors: KNeighborsClassifier
- GaussianProcessClassifier
- MLPClassifier
- SVC
1. Linear_model: LogisticRegression¶
1.1 Using feature set 1¶
In [3]:
# Parameter ranges
param_ranges = {
"penalty": ["l2"], # ["l1", "l2", "elasticnet"] certain combinations have conflict
"C": np.logspace(-4, 4, 4),
"solver": ["liblinear", "saga", "lbfgs", "newton-cg", "sag"],
"max_iter": [100, 200, 500],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=LogisticRegression, param_ranges=param_ranges, feature_set='1', cv_folds=5)
2024/09/25 08:22:17 INFO mlflow.tracking.fluent: Experiment with name 'LogisticRegression_1' does not exist. Creating a new experiment.
Hyperparameter combinations: 0%| | 0/60 [00:00<?, ?it/s]2024/09/25 08:22:17 WARNING mlflow.tracking.context.git_context: Failed to import Git (the Git executable is probably not on your PATH), so Git SHA is not available. Error: Failed to initialize: Bad git executable.
The git executable must be specified in one of the following ways:
- be included in your $PATH
- be set via $GIT_PYTHON_GIT_EXECUTABLE
- explicitly set via git.refresh(<full-path-to-git-executable>)
All git commands will error until this is rectified.
This initial message can be silenced or aggravated in the future by setting the
$GIT_PYTHON_REFRESH environment variable. Use one of the following values:
- quiet|q|silence|s|silent|none|n|0: for no message or exception
- warn|w|warning|log|l|1: for a warning message (logging level CRITICAL, displayed by default)
- error|e|exception|raise|r|2: for a raised exception
Example:
export GIT_PYTHON_REFRESH=quiet
Hyperparameter combinations: 100%|█████████████████████████████████████████████████| 60/60 [11:30:20<00:00, 690.33s/it]
1.2 Using feature set 2¶
In [4]:
# Parameter ranges
param_ranges = {
"penalty": ["l2"], # ["l1", "l2", "elasticnet"]
"C": np.logspace(-4, 4, 4),
"solver": ["liblinear", "saga", "lbfgs", "newton-cg", "sag"],
"max_iter": [100, 200, 500],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=LogisticRegression, param_ranges=param_ranges, feature_set='2', cv_folds=5)
2024/09/25 19:54:18 INFO mlflow.tracking.fluent: Experiment with name 'LogisticRegression_2' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████| 60/60 [1:11:46<00:00, 71.78s/it]
1.3 Using feature set 3¶
In [5]:
# Parameter ranges
param_ranges = {
"penalty": ["l2"], # ["l1", "l2", "elasticnet"]
"C": np.logspace(-4, 4, 4),
"solver": ["liblinear", "saga", "lbfgs", "newton-cg", "sag"],
"max_iter": [100, 200, 500],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=LogisticRegression, param_ranges=param_ranges, feature_set='3', cv_folds=5)
2024/09/25 21:07:43 INFO mlflow.tracking.fluent: Experiment with name 'LogisticRegression_3' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 60/60 [01:06<00:00, 1.11s/it]
1.4 Using feature set 4¶
In [6]:
# Parameter ranges
param_ranges = {
"penalty": ["l2"], # ["l1", "l2", "elasticnet"]
"C": np.logspace(-4, 4, 4),
"solver": ["liblinear", "saga", "lbfgs", "newton-cg", "sag"],
"max_iter": [100, 200, 500],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=LogisticRegression, param_ranges=param_ranges, feature_set='4', cv_folds=5)
2024/09/25 21:10:33 INFO mlflow.tracking.fluent: Experiment with name 'LogisticRegression_4' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 60/60 [08:13<00:00, 8.22s/it]
1.5 Using feature set 5¶
In [7]:
# Parameter ranges
param_ranges = {
"penalty": ["l2"], # ["l1", "l2", "elasticnet"]
"C": np.logspace(-4, 4, 4),
"solver": ["liblinear", "saga", "lbfgs", "newton-cg", "sag"],
"max_iter": [100, 200, 500],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=LogisticRegression, param_ranges=param_ranges, feature_set='5', cv_folds=5)
2024/09/25 21:20:26 INFO mlflow.tracking.fluent: Experiment with name 'LogisticRegression_5' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 60/60 [07:31<00:00, 7.52s/it]
1.6 Using feature set 6¶
In [8]:
# Parameter ranges
param_ranges = {
"penalty": ["l2"], # ["l1", "l2", "elasticnet"]
"C": np.logspace(-4, 4, 4),
"solver": ["liblinear", "saga", "lbfgs", "newton-cg", "sag"],
"max_iter": [100, 200, 500],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=LogisticRegression, param_ranges=param_ranges, feature_set='6', cv_folds=5)
2024/09/25 21:29:30 INFO mlflow.tracking.fluent: Experiment with name 'LogisticRegression_6' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 60/60 [06:56<00:00, 6.95s/it]
2. Naive_bayes: GaussianNB, BernoulliNB¶
BernoulliNB: Naive Bayes classifier for multivariate Bernoulli models.
2.1 Using feature set 1¶
In [9]:
# GaussianNB
# Parameter ranges
param_ranges = {
"var_smoothing": np.logspace(-9, -1, 30),
}
# Train and log the model
train_and_log_model(model_class=GaussianNB, param_ranges=param_ranges, feature_set='1', cv_folds=5)
2024/09/25 23:15:39 INFO mlflow.tracking.fluent: Experiment with name 'GaussianNB_1' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 30/30 [11:00<00:00, 22.02s/it]
In [10]:
# BernoulliNB
# Parameter ranges
param_ranges = {
"alpha": np.logspace(-3, 1, 5),
"binarize": [0.0, 0.5, 1.0],
"fit_prior": [True, False],
}
# Train and log the model
train_and_log_model(model_class=BernoulliNB, param_ranges=param_ranges, feature_set='1', cv_folds=5)
2024/09/25 23:32:04 INFO mlflow.tracking.fluent: Experiment with name 'BernoulliNB_1' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 30/30 [08:20<00:00, 16.70s/it]
2.2 Using feature set 2¶
In [11]:
# GaussianNB
# Parameter ranges
param_ranges = {
"var_smoothing": np.logspace(-9, -1, 30),
}
# Train and log the model
train_and_log_model(model_class=GaussianNB, param_ranges=param_ranges, feature_set='2', cv_folds=5)
2024/09/25 23:41:55 INFO mlflow.tracking.fluent: Experiment with name 'GaussianNB_2' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 30/30 [01:44<00:00, 3.47s/it]
In [12]:
# BernoulliNB
# Parameter ranges
param_ranges = {
"alpha": np.logspace(-3, 1, 5),
"binarize": [0.0, 0.5, 1.0],
"fit_prior": [True, False],
}
# Train and log the model
train_and_log_model(model_class=BernoulliNB, param_ranges=param_ranges, feature_set='2', cv_folds=5)
2024/09/25 23:45:14 INFO mlflow.tracking.fluent: Experiment with name 'BernoulliNB_2' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 30/30 [01:25<00:00, 2.84s/it]
2.3 Using feature set 3¶
In [13]:
# GaussianNB
# Parameter ranges
param_ranges = {
"var_smoothing": np.logspace(-9, -1, 30),
}
# Train and log the model
train_and_log_model(model_class=GaussianNB, param_ranges=param_ranges, feature_set='3', cv_folds=5)
2024/09/25 23:48:18 INFO mlflow.tracking.fluent: Experiment with name 'GaussianNB_3' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 30/30 [00:25<00:00, 1.17it/s]
In [14]:
# BernoulliNB
# Parameter ranges
param_ranges = {
"alpha": np.logspace(-3, 1, 5),
"binarize": [0.0, 0.5, 1.0],
"fit_prior": [True, False],
}
# Train and log the model
train_and_log_model(model_class=BernoulliNB, param_ranges=param_ranges, feature_set='3', cv_folds=5)
2024/09/25 23:50:25 INFO mlflow.tracking.fluent: Experiment with name 'BernoulliNB_3' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 30/30 [00:27<00:00, 1.10it/s]
2.4 Using feature set 4¶
In [15]:
# GaussianNB
# Parameter ranges
param_ranges = {
"var_smoothing": np.logspace(-9, -1, 30),
}
# Train and log the model
train_and_log_model(model_class=GaussianNB, param_ranges=param_ranges, feature_set='4', cv_folds=5)
2024/09/26 08:33:02 INFO mlflow.tracking.fluent: Experiment with name 'GaussianNB_4' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 30/30 [00:32<00:00, 1.10s/it]
In [16]:
# BernoulliNB
# Parameter ranges
param_ranges = {
"alpha": np.logspace(-3, 1, 5),
"binarize": [0.0, 0.5, 1.0],
"fit_prior": [True, False],
}
# Train and log the model
train_and_log_model(model_class=BernoulliNB, param_ranges=param_ranges, feature_set='4', cv_folds=5)
2024/09/26 08:35:13 INFO mlflow.tracking.fluent: Experiment with name 'BernoulliNB_4' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 30/30 [00:33<00:00, 1.13s/it]
2.5 Using feature set 5¶
In [17]:
# GaussianNB
# Parameter ranges
param_ranges = {
"var_smoothing": np.logspace(-9, -1, 30),
}
# Train and log the model
train_and_log_model(model_class=GaussianNB, param_ranges=param_ranges, feature_set='5', cv_folds=5)
2024/09/26 08:37:24 INFO mlflow.tracking.fluent: Experiment with name 'GaussianNB_5' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 30/30 [00:32<00:00, 1.08s/it]
In [18]:
# BernoulliNB
# Parameter ranges
param_ranges = {
"alpha": np.logspace(-3, 1, 5),
"binarize": [0.0, 0.5, 1.0],
"fit_prior": [True, False],
}
# Train and log the model
train_and_log_model(model_class=BernoulliNB, param_ranges=param_ranges, feature_set='5', cv_folds=5)
2024/09/26 08:39:35 INFO mlflow.tracking.fluent: Experiment with name 'BernoulliNB_5' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 30/30 [00:36<00:00, 1.22s/it]
2.6 Using feature set 6¶
In [19]:
# GaussianNB
# Parameter ranges
param_ranges = {
"var_smoothing": np.logspace(-9, -1, 30),
}
# Train and log the model
train_and_log_model(model_class=GaussianNB, param_ranges=param_ranges, feature_set='6', cv_folds=5)
2024/09/26 08:41:48 INFO mlflow.tracking.fluent: Experiment with name 'GaussianNB_6' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 30/30 [00:34<00:00, 1.15s/it]
In [20]:
# BernoulliNB
# Parameter ranges
param_ranges = {
"alpha": np.logspace(-3, 1, 5),
"binarize": [0.0, 0.5, 1.0],
"fit_prior": [True, False],
}
# Train and log the model
train_and_log_model(model_class=BernoulliNB, param_ranges=param_ranges, feature_set='6', cv_folds=5)
2024/09/26 08:43:59 INFO mlflow.tracking.fluent: Experiment with name 'BernoulliNB_6' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 30/30 [00:31<00:00, 1.06s/it]
3. Neighbors: KNeighborsClassifier¶
3.1 Using feature set 1¶
In [21]:
# KNeighborsClassifier
# Parameter ranges
param_ranges = {
"n_neighbors": np.arange(4, 21, 2),
"weights": ["uniform", "distance"],
"algorithm": ["auto", "ball_tree", "kd_tree", "brute"],
"leaf_size": np.arange(20, 51, 10),
"p": [1, 2]
}
# Train and log the model
train_and_log_model(model_class=KNeighborsClassifier, param_ranges=param_ranges, feature_set='1', cv_folds=5)
2024/09/26 08:46:07 INFO mlflow.tracking.fluent: Experiment with name 'KNeighborsClassifier_1' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████| 576/576 [7:43:54<00:00, 48.32s/it]
3.2 Using feature set 2¶
In [22]:
# KNeighborsClassifier
# Parameter ranges
param_ranges = {
"n_neighbors": np.arange(4, 21, 2),
"weights": ["uniform", "distance"],
"algorithm": ["auto", "ball_tree", "kd_tree", "brute"],
"leaf_size": np.arange(20, 51, 10),
"p": [1, 2]
}
# Train and log the model
train_and_log_model(model_class=KNeighborsClassifier, param_ranges=param_ranges, feature_set='2', cv_folds=5)
2024/09/26 16:31:40 INFO mlflow.tracking.fluent: Experiment with name 'KNeighborsClassifier_2' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████| 576/576 [1:06:58<00:00, 6.98s/it]
3.3 Using feature set 3¶
In [23]:
# KNeighborsClassifier
# Parameter ranges
param_ranges = {
"n_neighbors": np.arange(4, 21, 2),
"weights": ["uniform", "distance"],
"algorithm": ["auto", "ball_tree", "kd_tree", "brute"],
"leaf_size": np.arange(20, 51, 10),
"p": [1, 2]
}
# Train and log the model
train_and_log_model(model_class=KNeighborsClassifier, param_ranges=param_ranges, feature_set='3', cv_folds=5)
2024/09/26 17:40:14 INFO mlflow.tracking.fluent: Experiment with name 'KNeighborsClassifier_3' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████| 576/576 [11:14<00:00, 1.17s/it]
3.4 Using feature set 4¶
In [24]:
# KNeighborsClassifier
# Parameter ranges
param_ranges = {
"n_neighbors": np.arange(4, 21, 2),
"weights": ["uniform", "distance"],
"algorithm": ["auto", "ball_tree", "kd_tree", "brute"],
"leaf_size": np.arange(20, 51, 10),
"p": [1, 2]
}
# Train and log the model
train_and_log_model(model_class=KNeighborsClassifier, param_ranges=param_ranges, feature_set='4', cv_folds=5)
2024/09/26 17:53:09 INFO mlflow.tracking.fluent: Experiment with name 'KNeighborsClassifier_4' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████| 576/576 [18:33<00:00, 1.93s/it]
3.5 Using feature set 5¶
In [25]:
# KNeighborsClassifier
# Parameter ranges
param_ranges = {
"n_neighbors": np.arange(4, 21, 2),
"weights": ["uniform", "distance"],
"algorithm": ["auto", "ball_tree", "kd_tree", "brute"],
"leaf_size": np.arange(20, 51, 10),
"p": [1, 2]
}
# Train and log the model
train_and_log_model(model_class=KNeighborsClassifier, param_ranges=param_ranges, feature_set='5', cv_folds=5)
2024/09/26 18:13:26 INFO mlflow.tracking.fluent: Experiment with name 'KNeighborsClassifier_5' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████| 576/576 [17:14<00:00, 1.80s/it]
3.6 Using feature set 6¶
In [26]:
# KNeighborsClassifier
# Parameter ranges
param_ranges = {
"n_neighbors": np.arange(4, 21, 2),
"weights": ["uniform", "distance"],
"algorithm": ["auto", "ball_tree", "kd_tree", "brute"],
"leaf_size": np.arange(20, 51, 10),
"p": [1, 2]
}
# Train and log the model
train_and_log_model(model_class=KNeighborsClassifier, param_ranges=param_ranges, feature_set='6', cv_folds=5)
2024/09/26 18:32:22 INFO mlflow.tracking.fluent: Experiment with name 'KNeighborsClassifier_6' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████| 576/576 [18:52<00:00, 1.97s/it]
4. GaussianProcessClassifier¶
4.1 Using feature set 1¶
In [27]:
# Parameter ranges
param_ranges = {
# "kernel": None, # Note that the kernel’s hyperparameters are optimized during fitting.
"optimizer": ["fmin_l_bfgs_b"],
"n_restarts_optimizer": [0, 1, 5],
"max_iter_predict": [100, 200, 300],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=GaussianProcessClassifier, param_ranges=param_ranges, feature_set='1', cv_folds=5)
2024/09/26 18:52:55 INFO mlflow.tracking.fluent: Experiment with name 'GaussianProcessClassifier_1' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████████| 9/9 [14:53<00:00, 99.33s/it]
4.2 Using feature set 2¶
In [28]:
# Parameter ranges
param_ranges = {
# "kernel": None, # Note that the kernel’s hyperparameters are optimized during fitting.
"optimizer": ["fmin_l_bfgs_b"],
"n_restarts_optimizer": [0, 1, 5],
"max_iter_predict": [100, 200, 300],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=GaussianProcessClassifier, param_ranges=param_ranges, feature_set='2', cv_folds=5)
2024/09/26 19:10:02 INFO mlflow.tracking.fluent: Experiment with name 'GaussianProcessClassifier_2' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████████| 9/9 [02:14<00:00, 14.95s/it]
4.3 Using feature set 3¶
In [29]:
# Parameter ranges
param_ranges = {
# "kernel": None, # Note that the kernel’s hyperparameters are optimized during fitting.
"optimizer": ["fmin_l_bfgs_b"],
"n_restarts_optimizer": [0, 1, 5],
"max_iter_predict": [100, 200, 300],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=GaussianProcessClassifier, param_ranges=param_ranges, feature_set='3', cv_folds=5)
2024/09/26 19:14:00 INFO mlflow.tracking.fluent: Experiment with name 'GaussianProcessClassifier_3' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████████| 9/9 [00:24<00:00, 2.71s/it]
4.4 Using feature set 4¶
In [30]:
# Parameter ranges
param_ranges = {
# "kernel": None, # Note that the kernel’s hyperparameters are optimized during fitting.
"optimizer": ["fmin_l_bfgs_b"],
"n_restarts_optimizer": [0, 1, 5],
"max_iter_predict": [100, 200, 300],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=GaussianProcessClassifier, param_ranges=param_ranges, feature_set='4', cv_folds=5)
2024/09/26 19:16:04 INFO mlflow.tracking.fluent: Experiment with name 'GaussianProcessClassifier_4' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████████| 9/9 [00:36<00:00, 4.03s/it]
4.5 Using feature set 5¶
In [31]:
# Parameter ranges
param_ranges = {
# "kernel": None, # Note that the kernel’s hyperparameters are optimized during fitting.
"optimizer": ["fmin_l_bfgs_b"],
"n_restarts_optimizer": [0, 1, 5],
"max_iter_predict": [100, 200, 300],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=GaussianProcessClassifier, param_ranges=param_ranges, feature_set='5', cv_folds=5)
2024/09/26 19:18:20 INFO mlflow.tracking.fluent: Experiment with name 'GaussianProcessClassifier_5' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████████| 9/9 [00:36<00:00, 4.07s/it]
4.6 Using feature set 6¶
In [32]:
# Parameter ranges
param_ranges = {
# "kernel": None, # Note that the kernel’s hyperparameters are optimized during fitting.
"optimizer": ["fmin_l_bfgs_b"],
"n_restarts_optimizer": [0, 1, 5],
"max_iter_predict": [100, 200, 300],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=GaussianProcessClassifier, param_ranges=param_ranges, feature_set='6', cv_folds=5)
2024/09/26 19:20:38 INFO mlflow.tracking.fluent: Experiment with name 'GaussianProcessClassifier_6' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████████| 9/9 [00:34<00:00, 3.88s/it]
5. MLPClassifier¶
5.1 Using feature set 1¶
In [33]:
# Parameter ranges
param_ranges = {
"hidden_layer_sizes": [(16,32,64,32,16)],
"activation": ["logistic", "relu"],
"solver": ["adam"],
"learning_rate": ["constant", "adaptive"],
"max_iter": [200, 400, 500],
"tol": [1e-4],
"early_stopping": [True, False],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=MLPClassifier, param_ranges=param_ranges, feature_set='1', cv_folds=5)
2024/09/26 19:22:53 INFO mlflow.tracking.fluent: Experiment with name 'MLPClassifier_1' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|██████████████████████████████████████████████████| 24/24 [1:04:22<00:00, 160.94s/it]
5.2 Using feature set 2¶
In [34]:
# Parameter ranges
param_ranges = {
"hidden_layer_sizes": [(16,32,64,32,16)],
"activation": ["logistic", "relu"],
"solver": ["adam"],
"learning_rate": ["constant", "adaptive"],
"max_iter": [200, 400, 500],
"tol": [1e-4],
"early_stopping": [True, False],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=MLPClassifier, param_ranges=param_ranges, feature_set='2', cv_folds=5)
2024/09/26 20:29:09 INFO mlflow.tracking.fluent: Experiment with name 'MLPClassifier_2' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 24/24 [11:53<00:00, 29.73s/it]
5.3 Using feature set 3¶
In [35]:
# Parameter ranges
param_ranges = {
"hidden_layer_sizes": [(16,32,64,32,16)],
"activation": ["logistic", "relu"],
"solver": ["adam"],
"learning_rate": ["constant", "adaptive"],
"max_iter": [200, 400, 500],
"tol": [1e-4],
"early_stopping": [True, False],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=MLPClassifier, param_ranges=param_ranges, feature_set='3', cv_folds=5)
2024/09/26 20:42:45 INFO mlflow.tracking.fluent: Experiment with name 'MLPClassifier_3' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 24/24 [03:06<00:00, 7.76s/it]
5.4 Using feature set 4¶
In [36]:
# Parameter ranges
param_ranges = {
"hidden_layer_sizes": [(16,32,64,32,16)],
"activation": ["logistic", "relu"],
"solver": ["adam"],
"learning_rate": ["constant", "adaptive"],
"max_iter": [200, 400, 500],
"tol": [1e-4],
"early_stopping": [True, False],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=MLPClassifier, param_ranges=param_ranges, feature_set='4', cv_folds=5)
2024/09/26 20:47:35 INFO mlflow.tracking.fluent: Experiment with name 'MLPClassifier_4' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 24/24 [03:36<00:00, 9.03s/it]
5.5 Using feature set 5¶
In [37]:
# Parameter ranges
param_ranges = {
"hidden_layer_sizes": [(16,32,64,32,16)],
"activation": ["logistic", "relu"],
"solver": ["adam"],
"learning_rate": ["constant", "adaptive"],
"max_iter": [200, 400, 500],
"tol": [1e-4],
"early_stopping": [True, False],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=MLPClassifier, param_ranges=param_ranges, feature_set='5', cv_folds=5)
2024/09/26 20:52:58 INFO mlflow.tracking.fluent: Experiment with name 'MLPClassifier_5' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 24/24 [03:54<00:00, 9.76s/it]
5.6 Using feature set 6¶
In [38]:
# Parameter ranges
param_ranges = {
"hidden_layer_sizes": [(16,32,64,32,16)],
"activation": ["logistic", "relu"],
"solver": ["adam"],
"learning_rate": ["constant", "adaptive"],
"max_iter": [200, 400, 500],
"tol": [1e-4],
"early_stopping": [True, False],
"random_state": [42]
}
# Train and log the model
train_and_log_model(model_class=MLPClassifier, param_ranges=param_ranges, feature_set='6', cv_folds=5)
2024/09/26 20:58:37 INFO mlflow.tracking.fluent: Experiment with name 'MLPClassifier_6' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████████| 24/24 [03:34<00:00, 8.94s/it]
6. SVC¶
6.1 Using feature set 1¶
In [39]:
# Parameter ranges
param_ranges = {
'kernel': ['linear','poly', 'rbf', 'sigmoid'],
'C': [0.1, 1, 5, 10, 20], # Regularization parameter.
'gamma': ['scale', 'auto', 0.01, 0.1, 1], # Kernel coefficient for ‘rbf’, ‘poly’ and ‘sigmoid’.
}
# Train and log the model
train_and_log_model(model_class=SVC, param_ranges=param_ranges, feature_set='1', cv_folds=5)
2024/09/26 21:03:57 INFO mlflow.tracking.fluent: Experiment with name 'SVC_1' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|████████████████████████████████████████████████| 100/100 [3:30:31<00:00, 126.31s/it]
6.2 Using feature set 2¶
In [40]:
# Parameter ranges
param_ranges = {
'kernel': ['linear','poly', 'rbf', 'sigmoid'],
'C': [0.1, 1, 5, 10, 20], # Regularization parameter.
'gamma': ['scale', 'auto', 0.01, 0.1, 1], # Kernel coefficient for ‘rbf’, ‘poly’ and ‘sigmoid’.
}
# Train and log the model
train_and_log_model(model_class=SVC, param_ranges=param_ranges, feature_set='2', cv_folds=5)
2024/09/27 00:36:15 INFO mlflow.tracking.fluent: Experiment with name 'SVC_2' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|█████████████████████████████████████████████████| 100/100 [1:22:11<00:00, 49.32s/it]
6.3 Using feature set 3¶
In [41]:
# Parameter ranges
param_ranges = {
'kernel': ['linear','poly', 'rbf', 'sigmoid'],
'C': [0.1, 1, 5, 10, 20], # Regularization parameter.
'gamma': ['scale', 'auto', 0.01, 0.1, 1], # Kernel coefficient for ‘rbf’, ‘poly’ and ‘sigmoid’.
}
# Train and log the model
train_and_log_model(model_class=SVC, param_ranges=param_ranges, feature_set='3', cv_folds=5)
2024/09/27 02:00:04 INFO mlflow.tracking.fluent: Experiment with name 'SVC_3' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████| 100/100 [02:24<00:00, 1.44s/it]
6.4 Using feature set 4¶
In [42]:
# Parameter ranges
param_ranges = {
'kernel': ['linear','poly', 'rbf', 'sigmoid'],
'C': [0.1, 1, 5, 10, 20], # Regularization parameter.
'gamma': ['scale', 'auto', 0.01, 0.1, 1], # Kernel coefficient for ‘rbf’, ‘poly’ and ‘sigmoid’.
}
# Train and log the model
train_and_log_model(model_class=SVC, param_ranges=param_ranges, feature_set='4', cv_folds=5)
2024/09/27 02:04:10 INFO mlflow.tracking.fluent: Experiment with name 'SVC_4' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████| 100/100 [03:23<00:00, 2.03s/it]
6.5 Using feature set 5¶
In [43]:
# Parameter ranges
param_ranges = {
'kernel': ['linear','poly', 'rbf', 'sigmoid'],
'C': [0.1, 1, 5, 10, 20], # Regularization parameter.
'gamma': ['scale', 'auto', 0.01, 0.1, 1], # Kernel coefficient for ‘rbf’, ‘poly’ and ‘sigmoid’.
}
# Train and log the model
train_and_log_model(model_class=SVC, param_ranges=param_ranges, feature_set='5', cv_folds=5)
2024/09/27 02:09:18 INFO mlflow.tracking.fluent: Experiment with name 'SVC_5' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████| 100/100 [03:19<00:00, 2.00s/it]
6.6 Using feature set 6¶
In [44]:
# Parameter ranges
param_ranges = {
'kernel': ['linear','poly', 'rbf', 'sigmoid'],
'C': [0.1, 1, 5, 10, 20], # Regularization parameter.
'gamma': ['scale', 'auto', 0.01, 0.1, 1], # Kernel coefficient for ‘rbf’, ‘poly’ and ‘sigmoid’.
}
# Train and log the model
train_and_log_model(model_class=SVC, param_ranges=param_ranges, feature_set='6', cv_folds=5)
2024/09/27 02:14:19 INFO mlflow.tracking.fluent: Experiment with name 'SVC_6' does not exist. Creating a new experiment. Hyperparameter combinations: 100%|███████████████████████████████████████████████████| 100/100 [03:06<00:00, 1.86s/it]
